Live Speech Demos
We have provided a number of pre-sampled sound recordings below. These recordings give you an idea of how the qualtiy of speech varies by sampling rate. Higher frequencies produce the best playback sound. But, a high sampling rate has a corresponding cost since it requires greater amounts of memory for storage of data. When selecting a product you would have to compare the cost as a function of the sound quality (i.e., sampling frequency), amount of memory required for the application, features, etc.

ADPCM Algorithms

.wav files

These 16-bit wave files were resolved from 4-bit Oki ADPCM which are representative of the same quality at the same given sampling frequency. The file size of the original ADPCM files is 1/4 that of the expanded wave files below.

.aiff files

Select the appropriate file to execute playback in real-time.

The following formula will help you calculate the necessary amount of memory storage needed in your application.

Amount of memory required in the application (in bits) = The amount of recording time required in the application (seconds) x Data rate* x (BPS - Bits per second)
*Data rate = Sampling frequency (fsamp) x data format (3/4 bit ADPCM)

As an example, assume that you require 32 seconds of total recorded speech in your application and you wish to record at an 8 KHz sampling rate.
Size of memory required = 32 secs x (8 KHz x 4 bits) = 1,024,000 bits (OR)
32 seconds x 32 KBps (8 KHz fsamp x 4 bit ADPCM) = 1,024,000 bits or 128 Kbytes.

By knowing the total memory requirements of your application, you can refer to the Product Selection Guide to determine the correct speech device for your application and whether to use internal or external memory, based upon those requirements. For more information, see our Play-only Series and Record/Play Series tables.

Click here if you have a queste 'ion which you would like us to answer.